As is known, remote sensing is the science and art of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area, or phenomenon under investigation. Based on the wavelength in which the system works, remote sensing is principally categorized into two different groups, i.e., optical and microwave.
Optical remote sensing uses visible and infrared waves while microwave remote sensing uses radio waves.
As a microwave remote sensing system, a Synthetic Aperture Radar (SAR) system comprises a radar signal transmitter and a radar signal receiver operating on a movable platform, such as an aeroplane or satellite, and a remote processing station connected over a radio channel to the movable platform.
The transmitter sends a radar signal into a monitored ground or sea area, and the receiver receives the radar echo back-scattered by the area, and transmits it to the remote processing station, which processes the radar echo to obtain a two-dimensional map of the monitored area. The transmitted radar signal comprises a succession of microwave-band electromagnetic pulses modulated by linear frequency, or so-called CHIRP, modulation and transmitted at regular time intervals.
The processing station coherently combines the radar echoes corresponding to the transmitted pulses to obtain high-azimuth-resolution maps of extensive areas using relatively small transmitting antennas. CHIRP pulse modulation, on the other hand, provides for achieving high resolution perpendicular to the azimuth direction.
In detail, a SAR system illuminates a scene with microwaves by means of the transmitter, and records both amplitude and phase of the back-scattered radiation by means of the receiver, making it a coherent imaging process. The received signal is sampled and converted into a digital image by the processing station.
In particular, field recorded at pixel x, denoted E(x), can be written as
      E    ⁡          (      x      )        =            ∑      s        ⁢                  a        ⁡                  (          s          )                    ⁢              ⅇ                  ⅈ          ⁢                                          ⁢                      φ            ⁡                          (              s              )                                          ⁢              h        ⁡                  (                      s            ,            x                    )                    where the summation ranges over scatterers s, a(s) and φ(s) are respectively amplitude and phase of the signal received from the scatterer s, and h(s,x) is instrument or point-spread function. Value of h(s,x) is near 1 when the scatterer s is in or near the resolving cell corresponding to the pixel x, and near zero otherwise. Assuming that the instrument function is translation-invariant, i.e., it does not depend on x, then it can be written as a one-parameter function h(s−x).
The square of the modulus of the field E(x) is called detected intensity I(x); the square-root of the intensity I(x) is called envelope or amplitude. This is not the same as the amplitude of the received signal a(s) because the received field E(x) is perturbed by the instrument function h(s,x). The amplitude of the received signal a(s) is called reflectivity, and its square is called surface cross-section.
In compare to optical remote sensing, SAR imaging has some advantages. First, as an active system, it is a day/night data acquisition system. Second, considering the behaviour of electromagnetic waves in the range of SAR wavelength, it can be seen that atmospheric characteristics such as cloud, light rain, haze, and smoke has little effect on the capability of a SAR system. This makes SAR as an all-weather remote sensing system. Last but not least, as SAR signals partially penetrate into soil and vegetation canopy, in addition to surface information, it can provide subsurface information too.
Unfortunately, in compare to optical remote sensing, SAR imaging has some disadvantages too. In particular, unlike optical images, SAR images are formed by coherent interaction of the transmitted microwave with targets. Hence, SAR imaging suffers from the effects of speckle noise which arises from coherent summation of the signals back-scattered by ground scatterers s distributed randomly within each pixel x. Thus, a SAR image appears more noisy than an optical image.
More in detail, the waves emitted by the transmitter travel in phase and interact minimally on their way to the target area. After interaction with the target area, these waves are no longer in phase because of the different distances they travel from targets, or single versus multiple bounce scattering. Once out of phase, the back-scattered waves can interact to produce light and dark pixels. This effect is known as speckle noise.
The speckle noise gives a grainy appearance to the SAR images, reduces SAR image contrast, and has a negative effect on texture based analysis. Moreover, as the speckle noise changes spatial statistics of the SAR images, it makes the classification process a difficult task to do.
FIG. 1 shows an example of a typical grainy SAR image.
For this reason, the speckle noise is normally suppressed by applying a speckle removal filter on the digital image before display and further analysis.
FIG. 2 shows the grainy SAR image of FIG. 1 filtered with a speckle removal filter.
As shown in FIG. 1, a SAR image is usually displayed as a gray scale image. The intensity I(x) of each pixel x represents the portion of microwave back-scattered by a corresponding target area on the ground which portion depends on a variety of factors: types, sizes, shapes and orientations of the scatterers s in the target area, moisture content of the target area, frequency and polarization of the radar pulses, as well as the incident angles of the radar beam. The pixel intensity values are often converted into a physical quantity called back-scattering coefficient or normalized radar cross-section, which is measured in decibel (dB) units with values ranging from +5 dB for very bright objects to −40 dB for very dark surfaces.
Interpreting a SAR image is not a straightforward task. It very often requires some knowledge about the ground conditions of the areas imaged. As a useful rule of thumb, the higher the back-scattered intensity, the rougher is the surface being imaged.
In detail, flat surfaces such as paved roads, runways or calm water normally appear as dark areas in a SAR image since most of the incident radar pulses are specularly reflected away. FIG. 3 shows schematically an example of specular reflection. In detail, as shown in the FIG. 3, a smooth surface 31 acts like a mirror for the incident radar pulse. Most of the incident radar energy is reflected away according to the law of specular reflection, i.e. the angle of reflection α′ is equal to the angle of incidence α. Very little energy is back-scattered to the SAR sensor.
On the contrary, a rough surface reflects the incident radar pulse in all directions. This phenomenon is called diffused reflection. In this case, part of the radar energy is scattered back to the radar sensor. The amount of energy back-scattered depends on the properties of the target on the ground. FIG. 4 shows schematically an example of diffused reflection in presence of a rough surface 41.
Therefore, calm sea surfaces appear dark in SAR images, while rough sea surfaces may appear bright, especially when the incidence angle is small and when the roughness of the sea surface is due to waves having wavelength shorter than the wavelength of the radar pulses.
Moreover, the presence of oil films smoothes out the sea surface. Under certain conditions when the sea surface is sufficiently rough, oil films can be detected as dark patches against a bright background.
Furthermore, trees and other vegetations are usually moderately rough on the wavelength scale. Hence, they appear as moderately bright features in the image. Tropical rain forests have a characteristic back-scattering coefficient of between −6 and −7 dB, which is spatially homogeneous and remains stable in time. For this reason, the tropical rain forests have been used as calibrating targets in performing radiometric calibration of SAR images.
Very bright targets may appear in a SAR image due to a phenomenon called corner-reflection or double-bounce effect. FIG. 5 shows schematically an example of double-bounce effect where the radar pulse bounces off a horizontal ground 51 towards a target 52, and then is reflected from one vertical surface 52a of the target 52 back to the SAR sensor. Examples of such targets are ships on the sea, high-rise buildings and regular metallic objects such as cargo containers. Built-up areas and many man-made features usually appear as bright patches in a SAR image due to the corner-reflection effect.
FIG. 6 shows a SAR image of an area of the sea near a busy port wherein many ships can be seen as bright spots due to the corner-reflection effect. The sea is calm, and hence the ships can be easily detected against the dark background.
Finally, brightness of areas covered by bare soil may vary from very dark to very bright depending on its roughness and moisture content. Typically, rough soil appears bright in the image. For similar soil roughness, the surface with a higher moisture content will appear brighter.
FIG. 7 shows schematically an example of radar pulse interaction with a dry soil 71, wherein some of the incident radar energy is able to penetrate into the soil surface 71a, resulting in a less back-scattered intensity.
FIG. 8 shows schematically an example of radar pulse interaction with a wet soil 81, wherein the large difference in electrical properties between water and air results in a higher back-scattered radar intensity.
FIG. 9 shows schematically an example of radar pulse interaction with a flooded soil 91, wherein the radar pulses are specularly reflected off the water surface 92, resulting in low back-scattered intensity. The flooded area will appear dark in the SAR image.